Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: IV Regression with many zeros in 1st stage


From   "Mark Schaffer" <M.E.Schaffer@hw.ac.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: IV Regression with many zeros in 1st stage
Date   Fri, 7 May 2004 18:07:28 +0100

Michael,

Date sent:      	Fri, 7 May 2004 12:35:04 -0400
From:           	mganz <mganz@hsph.harvard.edu>
To:             	statalist <statalist@hsphsun2.harvard.edu>
Copies to:      	mganz@hsph.harvard.edu, Sarah Williamson <swilliamson80@hotmail.com>
Subject:        	st: IV Regression with many zeros in 1st stage
Send reply to:  	statalist@hsphsun2.harvard.edu

> Dear Statalist colleagues,
> 
> I would like to estimate the following regression:
> 
> BD=aX+bS+e
> 
> where BD=birth defect and S is a measure of smoking (either 0/1 or number of 
> cigarettes).  I would like to instrument S with cigarette prices and other 
> variables (local 
> smoking policy variables, eg).  The problem, I think, is that > 50% of 
> respondents didn't 
> smoke.  Does affect the appropriateness of the IV approach?  Ie, can my 1st 
> stage 
> regression:
> 
> S = cZ+dP+e
> 
> be estimated if >50% of S is zero?

IV is a one-step estimator, and it doesn't matter if the "first-stage 
regression" is correctly specified or not because it's not being 
estimated as a structural equation.  All you need is for S to be 
correlated with the excluded instruments.  You can check this with 
the 1st-stage F stat.

A couple of additional comments:

- If you want to do IV as 2SLS and estimate the first stage by hand, 
you have to include *all* the exogenous variables.  That means in 
your case you would include X along with Z and P.

- ...but there's no need to do it in two stages by hand; just use 
ivreg (or ivreg2, the extended IV written by Kit Baum, Steve Stillman 
and myself, which will provide the 1st-stage F-stat automatically if 
you want it).

- One issue that you haven't raised is whether your birth defect 
equation should be linear.  If BD is a dichotomous, you might want to 
us Joe Harkness' ivprob (probit with endogenous regressors) instead.

Hope this helps.

--Mark


> 
> Any help would be greatly appreciated.
> 
> Thanks,
> Michael
> 
> ________________________________________
> Michael Ganz, MS, PhD
> Assistant Professor
> Dept. of Society, Human Development and Health
> Harvard School of Public Health
> mganz@hsph.harvard.edu
> http://www.hsph.harvard.edu/faculty/ganz
> Ph. 617-432-2382
> Fax 617-432-3755
> 
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/


Prof. Mark E. Schaffer
Director
Centre for Economic Reform and Transformation
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS  UK
44-131-451-3494 direct
44-131-451-3008 fax
44-131-451-3485 CERT administrator
http://www.som.hw.ac.uk/cert
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index